Microprocessor-Based "Intelligent" Machines in Patient Care National Institutes of Health
Consensus Development Conference Statement
October 17-19, 1979
This statement is more than five years old and is provided solely for historical purposes. Due to the cumulative nature of medical research, new knowledge has inevitably accumulated in this subject area in the time since the statement was initially prepared. Thus some of the material is likely to be out of date, and at worst simply wrong. For reliable, current information on this and other health topics, we recommend consulting the National Institutes of Health's MedlinePlus http://www.nlm.nih.gov/medlineplus/.
This statement was originally published as: Microprocessor-Based "Intelligent" Machines in Patient Care. NIH Consens Statement 1979 Oct 17-19;2(10):59-62.For making bibliographic reference to the statement in the electronic form displayed here, it is recommended that the following format be used: Microprocessor-Based "Intelligent" Machines in Patient Care. NIH Consens Statement Online 1979 Oct 17-19 [cited year month day];2(10):59-62.
An NIH Consensus Development Conference on "The Use of Microprocessor-Based, 'Intelligent' Machines in Patient Care" was held in Silver Spring, Maryland on October 17-19, 1979.
The NIH consensus development program brings together practicing physicians, biomedical research scientists, consumers, and others in an effort to reach general agreement on the safety and efficacy of a medical technology, whether it be a drug, device, or procedure.
Progressive improvements in the art of manufacturing integrated circuits have shrunk the size and cost of many electronics devices. Tens of thousands of transistors and related components now routinely fit on silicon chips only 1/4 inch square. Nowhere, perhaps, have these engineering achievements had more pronounced effects than in the development of very small and inexpensive computers. Some specially designed microcomputers are small enough to fit within other devices or to be implanted biologically. The relative inexpensiveness and miniaturization of microcomputers have already led to the development of many new medical technologies.
Among the most interesting technologies may be those resulting from the incorporation of microcomputers into familiar medical devices. Microcomputer-based intensive care monitors, for example, might soon be capable of making medical judgments about patients and, where time is critical, autonomously modifying the patients' treatments in response to changes in their needs. The use of such devices--intended to improve patient care--would, however, raise both technological and sociological questions:
- Would these devices truly be more reliable or cost-effective than manual methods? Would they lead to improved patient outcomes?
- What ethical-legal issues can arise in their development, testing, and marketing?
- What might be their long-term, adverse consequences? How could these ill-effects be avoided?
- How might one verify that the devices perform those functions--and only those--which their designers intended?
The NIH consensus conference on medical uses of microprocessors* was convened to allow practitioners, researchers, and others to attempt to seek agreement on these issues. By meeting during a formative stage in the technology's development they might thus be able to guide its development along lines which are most responsive to society's needs.
* Microprocessors are large-scale integrated circuits which serve as the central processing units of microcomputers. As such they are key components controlling the interpretation and execution of machine instructions.
The conference began with nine invited papers on technical and social aspects of the subject--including six on recent applications. There then followed separate consensus panels which discussed the issues above in regard to use of the technology in (1) medical history taking, physical examinations, laboratory testing, and special procedures; (2) diagnosis and decision making; and (3) treatment and monitoring.
(Note: During the summer of 1978, the Biomedical Engineering and Instrumentation Branch, DRS, of the National Institutes of Health conducted a series of four workshops on "The Use of Microprocessor- Based, 'Intelligent' Machines in Patient Care" in preparation for the NIH consensus development conference. Fifteen panelists from the research and practicing communities participated by discussing: (1) the technology's state of the art, (2) its prospects for future development, and (3) some of its potential legal, ethical, social, political, medical, and economic implications. Copies of its report, NIH Pub. No. 79-1852, are available from Henry S. Eden, M.D., Assistant to the Chief, Biomedical Engineering and Instrumentation Branch, DRS, Building 13, Room 3W-13, NIH, Bethesda, Maryland 20205.)
What follows is a brief summary of some of the points raised by the audience and panels.
Consensus Development Panel 1
- Most panel members felt that so basic a development as small, inexpensive computers could potentially affect society in more fundamental ways then just through applications in medical settings. (As an example, microprocessor-based assembly line automation could displace workers.)
- Panel members stressed that although the electronic components of microcomputer-based systems were often inexpensive, total costs for such systems could be high, due to the expense of software development and regulatory compliance.
- Panel 1 felt that there was a medical role for microcomputers in the physician's office. They considered it desirable in some cases to have the results of certain tests (e.g., phonocardiograms, pulmonary function, blood chemistry profiles) available prior to the patient's examination. It was suggested that computerized, office-based instruments could make it possible for allied health personnel to acquire such data easily, quickly and precisely.
- The majority of Panel 2 felt that a proper goal for the development of systems for diagnosis and decision making should be "to exceed physician capability, not just emulate it." It was further felt that many commonly encountered medical decisions are neither difficult, nor do they require computerization. (One panelist further felt that some work in the area of "artificial intelligence" for medicine was predicated upon a false assumption--a shortage of physicians to make medical decisions.) The group listed the following technologically feasible, if not essential, applications of computers for diagnosis and decision making:
As a caveat, the group recalled early and somewhat unrealized expectations for automated systems in diagnosis and decision making. They cited the fuzzy and multidimensional nature of many medical decision processes as causes of difficulty in converting physicians' thought processes to machine-usable routines.
- use in auditing simple, routine decisions.
- assisting with a limited number of selected, complex decisions.
- providing improved access to medical information, and
- communicating information more quickly (e.g., from a clinical laboratory to an office).
- Members of Panel 3 observed that the characteristics of microprocessors--significant computing power in small, inexpensive packages--offer tremendous advantages to designers of instruments for patient therapy and monitoring. The panel noted promising developments underway in the form of implantable wearable, and bedside devices which were programmable through the use of microcomputers. Although the panel felt that the technical sophistication of microcomputers, per se, was fairly advanced, it noted a lack of ways to make miniaturized sensors and effectors capable of working in vivo for extended periods. This knowledge gap was felt to be inhibiting the development of effective closed-loop systems.
- Ethical and legal questions surrounding the use of microcomputers in medicine were not felt to be unique. There have, for example, been ethical and legal precedents for most issues, involving larger traditional computers and non-microprocessor-based wearable and implantable devices. It was felt, however, that debate on some of the issues--such as informed consent--may be given new life when the number and diversity of computer applications in medicine increase due to the availability of small, inexpensive computers.
- There was general discussion on whether computers used to acquire patient histories would depersonalize the doctor-patient encounter. The panels suggested that computers could be used effectively to ascertain the patient's "data base", but that the patient's "problem history"-- with its emotional overlays--could still be best acquired directly by the physician.
- The panels saw a need for effective standards for microprocessor-based medical instruments, particularly standards developed by professional medical and engineering societies. They drew a distinction, however, between "standards" and "standardization", the latter being too prone to inhibit further innovation. Panel 3 particularly commented that "the innovation process is a delicate one and probably needs more cultivation than control". At the same time, the panels acknowledged the existence of multiple and conflicting standards which have occasionally made device interconnection difficult.
- Potential problems with the documentation and evaluation of software were also acknowledged. Unanswered questions which the panels identified were: When users modify the software of a device, is a new device created? Who then is responsible for certifying that the modified device is safe and effective? Who will repair such devices? How can there best be a resolution of the conflicting needs of manufacturers--who wish to keep their software proprietary--and the needs of physicians--who wish to fully understand the algorithms used in devices serving patients under their care.
- The members of Panel 3 called for greater efforts to develop methods of assessing medical technologies and selectively encouraging the dissemination of those which are most cost-effective. It was observed that in many areas of the health care system the normal feedback mechanisms of the marketplace do not effectively select out the most appropriate uses of technology. The panelists felt that greater efforts are needed to prepare physicians to play this role, both in rejecting ineffective technologies and accepting and using those that are effective.
- Most panelists cited a need for better mechanisms for transferring recently developed technologies a) between institutions and b) from academia to the commercial sector. It was felt that this role could be played in part by federal seed monies which would assist in technology transfer.
Use in Medical History Taking, Physical Examination, Laboratory Testing and Special Procedures:
Cesar A. Caceres, M.D. (Chairman)
Mark S. Frankel, Ph.D. (Rapporteur)
Daniel Scott Janick, M.D., M.P.H.
Elwin Marg, Ph.D.
Roger G. Mark, M.D., Ph.D.
Helmuth F. Orthner, Ph.D.
Chariklia T. Spiegel, M.D.
Consensus Development Panel 2
Use in Diagnosis and Decision Making:
Morris F. Collen, M.D. (Chairman)
William A. Hymen, Ph.D. (Rapporteur)
J. Barclay Adams, M.D., Ph.D.
Thomas G. Field, Jr.. J.D.
Stanton Glantz, Ph.D.
Albert H. Tech, Ph.D.
Kenneth E. Warner, Ph.D.
Consensus Development Panel 3
Use in Treatment and Monitoring:
Ronald S. Newbower, Ph.D. (Chairman)
Victor Klig (Rapporteur)
Roy Branson, Ph.D
Ernest G. Cravalho, Ph.D.
Nicholas DeClaris, Sc. D.
Robert E. Fischell
Daniel Graupe, Ph.D.
Byron McLees, M.D.
Willis J. Tompkins, Ph.D.
NIH Division of Research Services
Biomedical Engineering and Instrumentation Branch
Office of Medical Applications of Research